CP13/6 - CRD IV for Investment Firms - Financial Conduct Authority

FCA 2013/xx
(1) appropriate minimum standards that the rating system is expected to
reach are defined, together with reasoning behind the adoption of such
standards and that the factors considered when determining the tests
are clearly documented;
(2) an objective rank-ordering metric, measured using an appropriate time
horizon (eg, using ratings one year prior to default) or cohort
approach), such as Gini or Accuracy Ratio of 50% is achieved over
time;
(3) where there are sufficient defaults from different time periods the
discriminatory power is shown to have reached the appropriate
minimum standard over an extended time period (ie, longest period
possible, including most recent data); and
(4) any concentrations in ratings from the model are demonstrated to be
appropriate.
4.3.146 G The FCA expects that a firm will not be compliant with the validation
requirements unless it can demonstrate in respect of the calibration that:
(1) observed default rate versus PD is considered at grade level and across
a range of economic environments (ie, as long as period as possible);
(2) where the PD does not relate to a pure point-in-time estimate, either
the PD or the observed default rate is transformed such that
comparison between the two is meaningful. This transformation
should be consistent with the model philosophy and calibration
technique applied; and
(3) pre-defined tolerances for the degree of divergence, and the associated
actions for what should happen when they are not met, are set.
4.3.147 G The FCA also expects that a firm will not be compliant with the validation
requirements unless it can demonstrate that:
(1) appropriate stability metrics should be considered across a range of
economic environments (ie, longest period possible including most
recent data);
(2) the tolerances for the degree of divergence, and associated actions for
what should happen when they are not met, is pre-defined; and
(3) subsections of portfolios by characteristics affecting risk profile, and
therefore potentially model performance, are investigated. Such
subsections could include:
(a)
loan type (amortising/interest only);
(b)
(c)
(d)
degree of hedging;
building type; and
other factors such as non-SPV (special purpose vehicle)
lending in a predominately SPV lending book or vice versa
(see article 188 of the EU CRR).
Page 89 of 197